The Download: Claude’s inner workings and OpenAI’s “super app”

Researchers at Anthropic have discovered new insights into how AI language models like Claude actually work internally. This kind of transparency research helps us better understand and trust AI systems.